基于Bi-LSTM特征融合和FT-FSL的非侵入式负荷辨识

doi:10.16088/j.issn.1001-6600.2025030701

摘要/Abstract

摘要： 通过非侵入式负荷监测(non-intrusive load monitoring,NILM)对负荷能耗进行实时监测和数据分析,能够实现能源合理配置和精细化管理。为了提高负荷标注数据不足情况下NILM的负荷识别效果,本文提出一种基于Bi-LSTM特征融合和微调小样本学习(fine-tuned few-shot learning,FT-FSL)的新方法应用于NILM。首先,通过Bi-LSTM将加权像素电压-电流(voltage-current,V-I)图像特征和多维时频序列特征进行融合;然后,通过FT-FSL使负荷分类模型能够基于少量标注数据进行训练;最后,在PLAID数据集上与4种主流FSL方法(包括匹配网络、原型网络、关系网络和MAML)进行对比实验。结果表明,本文方法的准确率达到92.46%,与对比模型相比,分别提高12.21个百分点、4.18个百分点、5.90个百分点和9.04个百分点,验证了本文方法能够有效识别标注数据不足的负荷类型。

关键词: 非侵入式负荷监测, 负荷辨识, 小样本学习, Bi-LSTM, 微调

Abstract: Non-intrusive load monitoring (NILM) facilitates the rational energy allocation and fine-grained management using real-time load data monitor and analysis. To improve load identification performance in NILM under conditions of limited labeled data, this paper presents a novel method based on Bi-LSTM feature fusion and fine-tuned few-shot learning (FT-FSL). First, weighted pixel voltage-current (V-I) image features and multidimensional time-frequency sequence features are fused using Bi-LSTM feature fusion method. Then, FT-FSL is employed to enable the load classification model to be trained with only a small number of labeled samples. Finally, the proposed method is evaluated on the PLAID dataset and compared with four mainstream FSL approaches (Matching Network, Prototypical Network, Relation Network, and MAML). Experimental results show that the proposed method achieves an accuracy of 92.46%, outperforming the comparison models by 12.21, 4.18, 5.90, and 9.04 percentage points, respectively. These results demonstrate the effectiveness of the proposed approach in identifying load types with limited labeled data.

Key words: non-intrusive load monitoring, load identification, few-shot learning, Bi-LSTM, fine-tuning

中图分类号: TM714

张竹露, 李华强, 刘洋, 许立雄. 基于Bi-LSTM特征融合和FT-FSL的非侵入式负荷辨识[J]. 广西师范大学学报（自然科学版）, 2026, 44(1): 33-44.

ZHANG Zhulu, LI Huaqiang, LIU Yang, XU Lixiong. Non-intrusive Load Identification Based on Bi-LSTM Feature Fusion and FT-FSL[J]. Journal of Guangxi Normal University(Natural Science Edition), 2026, 44(1): 33-44.

参考文献

[1] 谭全伟, 薛贵军, 谢文举. 基于VMD和RDC-Informer的短期供热负荷预测模型[J]. 广西师范大学学报(自然科学版), 2024, 42(5): 39-51. DOI: 10.16088/j.issn.1001-6600.2023082702.
[2] 李云卿. 系统推进“十四五” 时期电力需求侧管理高质量发展[J]. 电力需求侧管理, 2021, 23(1): 1-3. DOI: 10.3969/j.issn.1009-1831.2021.01.001.
[3] GOPINATH R, KUMAR M, JOSHUA C P C, et al. Energy management using non-intrusive load monitoring techniques-State-of-the-art and future research directions[J]. Sustainable Cities and Society, 2020, 62: 102411. DOI: 10.1016/j.scs.2020.102411.
[4] 余贻鑫, 刘博, 栾文鹏. 非侵入式居民电力负荷监测与分解技术[J]. 南方电网技术, 2013, 7(4): 1-5. DOI: 10.13648/j.cnki.issn1674-0629.2013.04.004.
[5] ANGELIS G F, TIMPLALEXIS C, KRINIDIS S, et al. NILM applications: literature review of learning approaches, recent developments and challenges[J]. Energy and Buildings, 2022, 261: 111951. DOI: 10.1016/j.enbuild.2022.111951.
[6] YIN H, ZHOU K L, YANG S L. Non-intrusive load monitoring by load trajectory and multi-feature based on DCNN[J]. IEEE Transactions on Industrial Informatics, 2023, 19(10): 10388-10400. DOI: 10.1109/TII.2023.3240924.
[7] 韩林池, 高放, 赵子巍, 等. 基于多尺度卷积与Informer混合模型的非侵入式负荷监测方法[J]. 电力自动化设备, 2024, 44(3): 134-141. DOI: 10.16081/j.epae.202306017.
[8] LE TT H, HEO S, KIM H. Toward load identification based on the Hilbert transform and sequence to sequence long short-term memory[J]. IEEE Transactions on Smart Grid, 2021, 12(4): 3252-3264. DOI: 10.1109/TSG.2021.3066570.
[9] DU L, HE D W, HARLEY R G, et al. Electric load classification by binary voltage-current trajectory mapping[J]. IEEE Transactions on Smart Grid, 2016, 7(1): 358-365. DOI: 10.1109/TSG.2015.2442225.
[10] DE BAETS L, RUYSSINCK J, DEVELDER C, et al. Appliance classification using VI trajectories and convolutional neural networks[J]. Energy and Buildings, 2018, 158: 32-36. DOI: 10.1016/j.enbuild.2017.09.087.
[11] 王守相, 郭陆阳, 陈海文, 等. 基于特征融合与深度学习的非侵入式负荷辨识算法[J]. 电力系统自动化, 2020, 44(9): 103-110. DOI: 10.7500/AEPS20190625010.
[12] CHEN T, QIN H Y, LI X S, et al. A non-intrusive load monitoring method based on feature fusion and SE-ResNet[J]. Electronics, 2023, 12(8): 1909. DOI: 10.3390/electronics12081909.
[13] 杨舒惠, 黎静华, 韦善阳. 基于图像分类网络的非侵入式负荷辨识算法的运算成本优化[J]. 电力自动化设备, 2024, 44(1): 141-146. DOI: 10.16081/j.epae.202303045.
[14] KASELIMI M, DOULAMIS N, VOULODIMOS A, et al. Context aware energy disaggregation using adaptive bidirectional LSTM models[J]. IEEE Transactions on Smart Grid, 2020, 11(4): 3054-3067. DOI: 10.1109/TSG.2020.2974347.
[15] RAFIQ H, SHI X H, ZHANG H X, et al. Generalizability improvement of deep learning-based non-intrusive load monitoring system using data augmentation[J]. IEEE Transactions on Smart Grid, 2021, 12(4): 3265-3277. DOI: 10.1109/TSG.2021.3082622.
[16] LIU Y Y, ZHONG L, QIU J, et al. Unsupervised domain adaptation for nonintrusive load monitoring via adversarial and joint adaptation network[J]. IEEE Transactions on Industrial Informatics, 2022, 18(1): 266-277. DOI: 10.1109/TII.2021.3065934.
[17] GIDARIS S, KOMODAKIS N. Dynamic few-shot visual learning without forgetting[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Los Alamitos, CA: IEEE Computer Society, 2018: 4367-4375. DOI: 10.1109/CVPR.2018.00459.
[18] CHEN C, LI K L, WEI W, et al. Hierarchical graph neural networks for few-shot learning[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2022, 32(1): 240-252. DOI: 10.1109/TCSVT.2021.3058098.
[19] 曾武, 毛国君. 多尺度特征聚合的小样本学习方法[J]. 计算机工程与应用, 2023, 59(15): 151-159. DOI: 10.3778/j.issn.1002-8331.2204-0278.
[20] HAN Y H, GAO Q, CHEN H Q, et al. A few-shot learning method for nonintrusive load monitoring with V-I trajectory features[J]. IEEE Sensors Journal, 2024, 24(7): 11867-11877. DOI: 10.1109/JSEN.2024.3365132.
[21] LUO Q Q, YU T, LAN C F, et al. A generalizable method for practical non-intrusive load monitoring via metric-based meta-learning[J]. IEEE Transactions on Smart Grid, 2024, 15(1): 1103-1115. DOI: 10.1109/TSG.2023.3263814.
[22] CHEN W Y, LIU Y C, KIRA Z, et al. A closer look at few-shot classification[EB/OL]. (2020-01-12)[2025-05-08]. https://arxiv.org/abs/1904.04232. DOI: 10.48550/arXiv.1904.04232.
[23] MEDICO R, DE BAETS L, GAO J K, et al. A voltage and current measurement dataset for plug load appliance identification in households[J]. Scientific Data, 2020, 7: 49. DOI: 10.1038/s41597-020-0389-7.
[24] DE BAETS L, DHAENE T, DESCHRIJVER D, et al. VI-based appliance classification using aggregated power consumption data[C]//2018 IEEE International Conference on Smart Computing (SMARTCOMP).Piscataway,NJ:IEEE, 2018: 179-186. DOI: 10.1109/SMARTCOMP.2018.00089.
[25] ALSHAREEF S, MORSI W G. Application of wavelet-based ensemble tree classifier for non-intrusive load monitoring[C]//2015 IEEE Electrical Power and Energy Conference (EPEC).Piscataway,NJ:IEEE, 2015: 397-401. DOI: 10.1109/EPEC.2015.7379983.
[26] HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Los Alamitos, CA: IEEE Computer Society, 2016: 770-778. DOI: 10.1109/CVPR.2016.90.
[27] ZHOU H Y, ZHANG S H, PENG J Q, et al. Informer: beyond efficient transformer for long sequence time-series forecasting[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2021, 35(12): 11106-11115. DOI: 10.1609/aaai.v35i12.17325.
[28] NIE W Z, YAN Y,SONG D, et al. Multi-modal feature fusion based on multi-layers LSTM for video emotion recognition[J]. Multimedia Tools and Applications, 2021, 80(11): 16205-16214. DOI: 10.1007/s11042-020-08796-8.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed